CT Perfusion

CT Perfusion, By Dr. Houman Khosravani

Imaging Preamble:

This section is not a comprehensive or exhaustive look at computed tomography (CT) perfusion. There are many good sources to understand the physics underlying CT perfusion (CTP). Nonetheless herein I provide a framework to understand CTP such that as a practitioner you can begin to use this powerful tool to make hyperacute treatment decisions.

Overall - acute stroke imaging tries to accomplish several goals that are pertinent to the evaluation of the patient in the hyperacute setting. These include the following: Whether there is hemorrhage, whether there is an intra-arterial thrombus that could be a target for thrombolysis or thrombectomy, whether there is an ischemic core/already infarcted tissue, and lastly whether a penumbra of tissue exists that is ischemic but potentially salvageable (i.e. ischemic but not infarcted).

CT perfusion is well apt at addressing whether there is an infarct core or tissue that has surpassed the critical limit of ischemia - that is tissue that is ischemic but is not yet infarcted termed the “penumbra”. As neuronal perfusion is further compromised, ischemia evolves to become the infarct “core”.

In acute stroke imaging, four types of image series are acquired. These include plain CT head (non-contrast CT head, NCCT), followed by CT angiography (CTA) which traditionally spans the aortic arch to the vertex, followed by CT perfusion, and finally a post-contrast CT head. Certainly at most centers, CT angiography is reformatted to both coronal and sagittal planes in addition to the axial acquisition.

At our shop there are 4 CT image types that are acquired - each with important information:

  • Plain CT
  • Plain CT High-resolution
  • Multiphasic CTA
    • CTA
    • CTA Multiphase
    • Coronal
    • Sagittal
    • Axial
  • CT Perfusion

Multiphasic CT angiography:

Before we discuss CT a word about multiphasic CT. CTA can further be augmented by additional scans spanning only the cranium/intracranial vessels, with a time delay resulting in two additional head-only scans. These additional scans generate what is referred to as multiphase CT angiography. Multiphasic CT is used to assess the presence and amount of collaterals but also whether slow-moving blood is present. On additional phases of CTA (i.e. phases 2 and 3), the presumed normal side will transition quickly from arterial phase into the venous phase and there will be no contrast that is lingering in the arterial branches. However, if there is a large vessel occlusion, the side that is affected, will have contrast lingering on phase 2 and phase 3 suggesting increased transit time. Multiphasic CT does not replace CT perfusion, however it does contain some of the similar information parameters that can be obtained on perfusion imaging as well.

CT Perfusion:

Fundamentally, CT perfusion refers to how blood flows at the capillary level and this data is extracted through perfusion of dye, that is performed as a contrast injection after the conventional CT angiography.

Considerable variety exists and protocols for CT perfusion scanning, and processing of the perfusion maps. Recently standardization of tools to characterize perfusion have had a pivotal role in identifying thrombectomy candidates in the late window between 6-24 hours. Currently the algorithms utilized for this particular use case (in the DEFUSE3 and DAWN clinical trials) have used the RAPID software (iSchemia View) but there are other algorithms available as well that compute CTP maps.

Before going further, it is important to note that MRI perfusion also exists, however this is beyond the scope of this text. Advantages of CT perfusion include rapid acquisition, improved resolution, quantitative information, ease of repeatability (e.g. in conditions were repeat perfusion studies are required such as vasospasm in SAH). These advantages include limitations with regards to coverage (spatial across the head), the fact that ionizing radiation is used scanning a particular location multiple times, contrast risk and allergy, and need for post-processing. The latter has much improved with streamlined algorithms such as RAPID.

Framework for understanding CTP:

CT perfusion is described by several key parameters including: CBF (cerebral blood flow), CBV (cerebral blood volume), and MTT (mean transit time). Cerebral blood volume is defined as the total volume of blood, flowing through a volume of brain (ml/100g of brain tissue). CBF is defined by the volume of blood moving through a volume of the brain per unit time (ml/100g/min). MTT is defined as the average transit time of blood, for a given brain region measured in units of time (seconds or minutes). Core and penumbra have had different definitions over time. Previously core was defined as CBV volume, but the newer measures use CBF below a certain threshold is defined as the core. This is based on additional research that suggests that cerebral blood flow is more of an optimal parameter for assessing infarct core. Similarly mean transit time could be used to define a penumbra, and a more sensitive measure of that is Tmax. This quantity (Tmax, i.e. time to peak TTP) comes from the deconvolved tissue residue function.

CT perfusion is a dynamic measure and involves intravenous contrast administration, and this contrast is tracked with serial imaging during its "first pass" circulation through the brain tissue capillary bed. To calculate perfusion, there are a few assumptions that are made, one is that the perfusion tracer is not diffusible, metabolized or absorbed by the tissue. This is generally felt to be the case in the normal brain, however in the context of injury (including stroke), there is a breakdown of the blood brain barrier due to inflammation, and therefore this assumption does not fully hold true. Generally this results in overestimation of the CT CBV.

Determination of cerebral perfusion also relies on understanding the relation between arterial and venous enhancement. Underpinning this is mathematical formulations that deal with tracer kinetic theory (Fick Principle).

As the contrast agent passes through brain tissue, there is a transient increase in signal in the vessels and also the tissue in proportion to the amount of contrast in the blood vessels in that region. It is through measurement of this that time attenuation curves are generated, and for regions of interest in arterial and venous regions, and unaffected vessels that are perpendicular to the acquisition plane. Relative to these the perfusion maps are generated. The venous function is used to correct for some of the averaging (referred to as volume or signal averaging) effects. Technical issues include timing of contrast. Due to the complexity of acquisition, depending on the scanner being utilized, CT perfusion images may or may not cover the entire brain, and may be limited to a specific region.

Conceptually it is easy to think about cerebral blood flow as the ratio of cerebral blood volume moving across time, therefore, CBF = CBV/MTT. The mathematics and types of calculations performed (deconvolution vs. deconvolution-based algorithms) is beyond the scope of this text. However we will outline the conceptual framework that could be utilized to understand perfusion imaging. Furthermore, we will tie the key variables of CBF, CBV, and MTT to physiologic concepts that are at play during the initial phase of stroke management.

Some practical considerations and assumptions that CT perfusion imaging requires is that the patient has good cardiac output, is in normal sinus rhythm (certainly some patients are in atrial fibrillation who presented in acute stroke), absence of any proximal arterial stenosis. During the calculation of CT perfusion parameters, areas of interest (AOIs) have to be selected, and these form the underpinnings of the arterial and venous time density curves. Qualitative assessment of these density curves also provides information with regards to the efficacy of the contrast injection. Furthermore, CT perfusion imaging relies on some thresholds, that are computational in nature and at times predetermined, and the skin because of errors in identifying core and penumbra.

CT perfusion is most useful in the hemispheric infarcts, such as MCA territory, and in the case of scanners that are able to look at more caudal regions, brainstem and cerebellum. However it is important to note that most of the validated studies of CT perfusion are focused on the MCA territories. By corollary, smaller infarct such as lacunar infarcts are not well visualized on CT perfusion maps. Furthermore other conditions such as seizures may result in maps that show the opposite of his stroke, such as hyperemia with increased perfusion due to the ictal state. However, depending on the timing of seizure activity, and the nature of the seizure itself, the appearance of CT perfusion is not always in one consistent manner that identifies underlying seizure activity.

In areas of the brain where an infarct has completed, there is decreased cerebral blood flow, and decreased cerebral blood volume with ongoing increased mean transit time. This condition or rather endpoint is a pattern consistent with loss of sufficient perfusion to neurons, and therefore can be thought of as a reversible loss of function corresponding with neuronal death. Certainly there are limits to CT perfusion, and tissue identified as infarct core may have aspects that are still salvageable.

Furthermore, after ischemia has occurred, sometimes there is hyperperfusion of the tissue, so-called luxury perfusion, that sufficiently restored blood flow to an area of infarct, and this can cause underestimation of the actual infarct core. At times when there is a vessel occlusion and recanalization, both cerebral blood volume and cerebral blood flow are increased within the distal territory, compared to the contralateral brain, and this time artificially the contralateral may have an appearance of ischemia but this is not a true identification but rather erroneous in nature. Also, CT perfusion maps are qualitatively observed, and many software packages utilize color gradients in order to demonstrate numerical ranges, however more recent soft for utilizes calculations of volume in millilitres, for both infarct core and penumbra which make them easier to appreciate and understand in the acute setting. Furthermore the concept of mismatch refers to a penumbra that is larger than the core, and once again numerical values are now being automatically assigned in order to appreciate that there is a mismatch.

In CT scanners where entire brain coverage is not possible, user defined regions of interest for scanning are important in answering the clinical question for example whether the location of occlusion is in the posterior or anterior circulation. In most centers a balanced series of slices is selected to adequately cover the MCA regions, with some caudal coverage.

Using the framework and the algebraic relationship between CBF = CBV/MTT, during the initial phase of tissue hyperperfusion, cerebral blood flow is either maintained or just slightly decreased because of distal vasodilation that increases cerebral blood volume in proportion to the elevated mean transit time - the net result is a slight decrease in CBF. However as time progresses, maximal vasodilatation has occurred and CBV can no longer increase in magnitude while MTT is prolonged. This is furthermore compounded by tissue injury at the cellular level which ultimately then results in a marked decreased in CBV, resulting in fairly marked decrease in CBF. At this juncture, the CBF falls below the ischemic threshold and thus tissue begins to infarct.

CTP requires thinking - considerations and pitfalls:

With the advent of algorithmic computation of CT perfusion maps, and there readily available nature during a code stroke, there can be a tendency to use these maps to make a clinical decision without thinking more deeply about what the perfusion maps actually show. During a code stroke, one must stay vigilant and attentive to the details of what CT perfusion is showing in the context of the patient but also about what the analysis actually shows.

As noted earlier, CT perfusion makes a series of assumptions, both at the technical level and also about the patient's hemodynamics. Furthermore, there are other technical factors such as movement of the patient, and the quality and timing of the contrast injection. If all of these are minimized and optimized, to understand a CT perfusion maps one still needs to deploy some understanding of what is actually being shown on the perfusion maps.

From a simplistic perspective, the amount of infarct core must be compared to the amount of penumbra, and if these 2 regions are similar in size, the perfusion abnormality is termed "matched" -meaning that a large portion of penumbra has already infarcted, and that there is less likely of a benefit for revascularization with either TPA and/or mechanical thrombectomy.

However if there is significant mismatch, then there may be salvageable brain. In general, perfusion is not reliable within the first 4-6 hours of stroke onset, and therefore the greatest utility of perfusion is in the late time windows, specifically the 6-24 H epoch. For this time window, there are specific criteria that are utilized in order to determine whether the region of available mismatch qualifies for revascularization. In general, the mismatch ratio of penumbra to core (expressed in ml for volume) must be greater than 1.8.

Furthermore more streamlined algorithms such as RAPID, not only demonstrate an area of core/infarct as CBF < 30%, and penumbra expressed as Tmax > 6 sec., but also provide additional maps with gradations of both cerebral blood flow and Tmax, such that the provider can consider various thresholds, in the context of the individualized patient, in order to assess whether the perfusion map is accurately reflecting a perfusion abnormality.

CT Perfusion parameters during hyperacute stroke:

DAWN / DEFUSE 3 Criteria - to be used for > 6hr for EVT Consideration

RAPID Training (www.irapid.com/training-rapid)

Late EVT Window, Greater than 6 hours (6-24Hrs)

DEFUSE3 criteria

  • ICA or MCA-M1 occlusion (carotid occlusions can be cervical or intracranial; with or without tandem MCA lesions) by MRA or CTA AND
  • Target Mismatch Profile on CT perfusion or MRI (ischemic core volume is < 70 ml (CBF<30), mismatch ratio is >/= 1.8 and mismatch volume* is >/= 15 ml) Alternative neuroimaging inclusion criteria (if perfusion imaging or CTA/MRA is technically inadequate): If CTA (or MRA) is technically inadequate:
  • Tmax>6s perfusion deficit consistent with an ICA or MCA-M1 occlusion AND Target Mismatch Profile (ischemic core volume is < 70 ml, mismatch ratio is >1.8 and mismatch volume is >15 ml as determined by RAPID software)

DAWN criteria (core measured CBF<30)

  • 0-<21 cc core infarct and NIHSS ≥ 10 (and age ≥ 80 years old)
  • 0-<31 cc core infarct and NIHSS ≥ 10 (and age < 80 years old)
  • 31 cc to <51 cc core infarct and NIHSS ≥ 20 (and age < 80 years old)

Lastly a word of caution. ALWAYS ensure you look at NCCT to assess hypo-density and correlate this with CBF core recalling that you may have a sub-acute infarct with luxury perfusion (therefore CBF>30 i.e. not showing as core on threshold to CBF map).


Graphic/table reference: Stroke, G. A., 2018. (n.d.). Use of Imaging to Select Patients for Late Window Endovascular Therapy. Am Heart Assoc. http://doi.org/10.1161/STROKEAHA.118.021011.)

RAPID Software Resources

Practical_approach_to_RAPID.pdf